This session addressed issues related to the development of thresholds. Should thresholds vary for different population groups? How to capture variability of risk across populations, including geographic variations in exposure to medical care economic risk, and vulnerability of population groups by insurance status, age, income, chronic health conditions? How to update the thresholds?
The presentations covered three topic areas affecting the development of thresholds:
- Geographic variations in exposure to medical care economic risk.
- Trends in persistent financial burden of medical out-of-pocket expenditures resulting from chronic health conditions.
- Trends in insurance coverage and their impact on medical out-of-pocket expenses 6.
TRACKING GEOGRAPHIC VARIATIONS IN EXPOSURE TO MEDICAL CARE ECONOMIC RISK: MOVING BEYOND ONE NATIONAL ESTIMATE
Sara Collins (The Commonwealth Fund) opened her presentation by observing that high out-of-pocket medical costs are an increasing problem for insured as well as uninsured people. She referred to an analysis relevant to the topic in which she participated on the numbers of underinsured and the trends over time (Schoen et al., 2011). The authors found that national out-of-pocket spending on health care services, not counting premiums,
rapidly rose in both percentages and numbers over the period 2003-2010. Their analysis shows growing numbers of working-age adults who have such high out-of-pocket costs relative to their income that they are underinsured. The authors found that an estimated 29 million people, 16 percent of the adult population, were underinsured, up from 16 million, or about 9 percent, in 2003.
She said that these measures are based on the The Commonwealth Fund’s Biennial Health Insurance Survey with a dual frame landline and cell phone sample of 4,000 nonelderly persons. The underinsured were defined as spending 10 percent or more of their income on medical expenses, or 5 percent or more if their incomes were under 200 percent of poverty, or deductibles equal to 5 percent or more of income.
The Affordable Care Act (ACA) will significantly expand and improve health insurance coverage with the expansion in Medicaid, income-related premium tax credits, basic health care plans if states opt to implement one, and reduced cost-sharing to limit the risk of high out-of-pocket costs and also enable timely access to health care.
There are remaining risks, however. Households with low and moderate incomes may still be at risk of high costs depending on the design of health plans and the choices people make among health plans available to them, state health care spending growth, exemptions in the law of certain plans, grandfathered plans, and self-insured plans. Also, state implementation decisions, especially with respect to the exchanges, and state enforcement of insurance market reforms may affect the risk of high costs.
The Insurance Affordability Programs that will be available under the ACA beginning in 2014—Medicaid, the Children’s Health Insurance Program, the basic health plan, premium tax credits, and the cost-sharing protections for qualified health plans in the exchanges—will go a long way toward reversing the trends on underinsurance and on the uninsured, particularly for people with incomes under 250 percent of poverty for whom the protections are the greatest.
But even for higher income households, the limits on out-of-pocket spending, market reforms against underwriting the essential benefit package, and guaranteed issue and no preexisting condition exclusions will also offer new protections. Schoen and her colleagues estimated that the ACA could reduce the number of underinsured adults by 70 percent.
Some work published in a Commonwealth Fund issue brief of the ACA’s insurance expansions found that 90 percent of households with median out-of-pocket spending would have sufficient room in their budgets for both premiums and out-of-pocket costs after full ACA implementation (Gruber and Perry, 2011). Using data from the Medical Expenditure Panel Survey (MEPS) and the Consumer Expenditure Survey, they established a standard for necessities and then assessed whether there was sufficient
additional income in budgets to pay for both health insurance and out-of-pocket costs.
They used the family economic self-sufficiency standard, which considers such necessities as child care, food, housing, taxes, transportation, and miscellaneous costs, which are defined as 10 percent of other costs. They found risks in the analysis: people with high spending in any given year, such as people with chronic health problems or catastrophic accidents, 25 percent of those with incomes between 200 and 250 percent of poverty would not have had room in their budgets for the premiums and the out-of-pocket costs.
As income rises past 200 percent of poverty, cost exposure also rises. Premium cost-sharing increases from 6.3 to 8.5 percent of income as income moves from 200 percent of poverty to 250 percent of poverty.
Gruber and Perry also found differences by states. Higher shares of people in states with higher cost of living would not have had room in their budgets for premiums and out-of-pocket costs. This is exacerbated for people living in states with a high cost of living who also had high health care spending: more than 30 percent of people in this group with incomes between 200 and 250 percent of poverty did not have room in their budgets for premiums and out-of pocket costs. The March 2010 Current Population Survey (CPS) asked about total out-of-pocket expenditures, including premiums, and enabled estimates of risk at the state level, an exciting development in terms of tracking and understanding what has been going on across states. The CPS asks about total out-of-pocket costs for medical care services in 2009, including premiums and costs reimbursed by insurance. For households with more than one member, the data files aggregate spending for each family member in total family expense. Preliminary estimates based on one year of data in 2010 indicate significant variation in the percentage of families with high out-of-pocket spending across states. Tracking trends in out-of-pocket costs nationally and by state for those insured, uninsured, and by poverty status will help inform reform implementation and future policies.
Collins briefly explained the data and method used in the analysis she conducted with Bhaven Sampat, Cathy Schoen, and Nicholas Tilipman, from Columbia University. They used the new out-of-pocket spending measure in the 2010 CPS to analyze out-of-pocket cost burden at the household/family level, also using a measure of family income. The CPS asks about total out-of-pocket costs for medical services in 2009, including premiums and costs reimbursed by insurance. For households with more than one member, the data files aggregate spending for each family member in total family expense. They classified a household as insured if all members in the family are insured. Uninsured families are families in which everyone is uninsured or some members are uninsured. They defined out-of-pocket
spending risk, or thresholds, as families that spent 10 percent or more of their income annually on medical needs, or people who spent 5 percent of their income, if their incomes were under 200 percent of poverty. The results show the percentage of families and total counts of people in families with high medical care expenses compared with their income.
Tilipman and Sampat, in their analysis of the 2010 CPS for the Commonwealth Fund, examined two threshold measures of out-of-pocket spending and premiums: 10 percent or more of income and 5 percent or more if income was under 200 percent of poverty.1 They found that, among all families at the 10 percent or more threshold, 13 percent of families spent 10 percent or more of their income on out-of-pocket expenses. If the 5 percent threshold is added, that jumped up to 17 percent. Among insured families, 11 percent of families spent 10 percent or more of their income on out-of-pocket costs. The percentage increases to 15 percent of families if the 5 percent threshold is added. Among uninsured families, the risks were clearly the highest, with nearly a quarter of these families experiencing high out-of-pocket spending relative to their low incomes, using the combined 10 percent and 5 percent threshold—a big jump when the lower threshold is added.
Families most at risk were those with low incomes, Collins said. Nearly 40 percent of families under 133 percent of poverty had high out-of-pocket costs, and 31 percent of those with incomes between 133 and 199 percent of poverty also had high out-of-pocket spending relative to their incomes.
When they looked at just insured families, 37 percent of those with incomes under 133 percent of the poverty level and nearly 40 percent of those with incomes under 133-139 percent of the poverty level had high out-of-pocket medical care costs and premiums. It shows a combination of low income and also poor coverage options at that income level, less employer coverage among low-income groups, and people in employer plans with high cost-sharing relative to lower paying jobs.
Collins observed that a big advantage of the 2010 data is being able to look at differences across states in medical care cost spending. Families in the Southeastern states were most at risk of spending large amounts of their income on out-of-pocket health care costs, amounting to nearly a quarter of the population in those states spending 10 or 15 percent of their income on out-of-pocket medical care costs. Some of the highest spending states were Alabama, Arizona, Arkansas, Indiana, Kentucky, Mississippi, and Tennessee. Focusing just on insured families, Mississippi and Arkansas had had the highest rates of high out-of-pocket spending, again reflecting
1 Analysis of the 2010 Current Population Survey, U.S. Census Bureau, by Nicholas Tilipman and Bhaven Sampat of Columbia University for The Commonwealth Fund.
a combination of low income and poor coverage. New England and the Midwest saw slightly lower rates.
She cautioned, however, that it is important to keep in mind in looking at these rankings that some of these levels are separated by 1 percentage point differences. In all states, families with lower income are most at risk due to higher rates of uninsured and also less protective coverage.
To summarize, 17 percent of families, about 44 million people, had high out-of-pocket costs in 2009 relative to their income. Most at risk were low-income households; nearly 40 percent of insured families under 200 percent of the federal poverty level had high out-of-pocket costs.
High out-of-pocket costs varied dramatically by state, ranging from 12 to 24 percent of families.
Collins observed that the ACA reforms beginning in 2014, with Medicaid expansion, premium tax credits, and lower cost sharingfor qualified health plans in the exchanges, the essential benefit package, and market reforms, should lead to a dramatic reduction in the share of families with high out-of-pocket costs as a share of their income both nationally and across states. But risks include ongoing risks of rapid health care cost growth compared with income, families with chronic illness, the design of benefit plans and the state implementation of the exchanges, enrollment coordination between coverage options, the pace of implementation, and the exemptions of health plans from the law. She concluded by saying that all of this suggests a need to monitor the law over time, at the state and national levels.
HIGH MEDICAL CARE COST BURDENS AMONG NONELDERLY ADULTS WITH CHRONIC CONDITIONS
Peter Cunningham (Center for Studying Health Systems Change) presented findings related to trends in out-of-pocket spending and high financial burden, how they have changed over time, how they differ for different population groups, and what they imply for affordability thresholds. When one thinks about affordability thresholds for medical care, the question is whether they should be different for people with chronic conditions or people with health conditions that require high expenditures. Clearly, people with health problems use more health care, and they spend a lot more on health care than people with fewer or no health problems. But does that necessarily mean that they should have a different affordability threshold?
For example, if it is determined that health care should be affordable up to, say, 5 percent of a family’s income, is there a reason why that should be different for people with chronic conditions? It could simply mean that people with chronic conditions are going to meet that threshold more often. To make the case that there should be a different threshold for people with
chronic conditions or other health problems, one has to argue that affordability is different in some way for persons with chronic conditions other than simply their high level of spending.
Cunningham pointed out that the ACA does recognize different affordability thresholds based on income as a way of calculating the premium subsidies and even subsidies for cost-sharing. Although he said he is not advocating that there should be a different threshold for people with chronic conditions, such an argument could be made based on the findings of his research, which he presented. The research shows that high medical cost burdens for people with chronic conditions tend to persist over time, and these can often lead to greater accumulated debt, which in any given year creates more financial pressures than simply what they spend in that particular time period.
He stated that most of the findings in his presentation are based on 2008 data from MEPS, which were the most recently available data at the time of the workshop. Most data on expenditures and high medical cost burdens are based on annual estimates, and they are retrospective. But he has also used the panel component of MEPS to look at the persistence of high burden over a 2-year period. High financial burden is defined as out-of-pocket spending for both premiums and services that exceed 10 percent of family income, a definition that is consistent with his past work. Before-tax incomes are used, and assets are not included. Also, the sample is limited to nonelderly adults.
Health Insurance Coverage by Health Conditions
Using the conditions file and coding by the International Classification of Diseases, ninth revision (ICD-9), and the clinical classification codes that are in the MEPS, nonelderly adults were classified on the basis of their insurance status, whether they had any reported conditions during the year, whether they had acute conditions only, and whether they reported one, two, or three or more chronic conditions. The data show that nonelderly adults with multiple health problems were not necessarily at greater risk for high financial burden because of lack of coverage. In fact, among all the conditions looked at, they had the lowest uninsured rates of all the groups. In fact, people with no conditions had the highest uninsured rates, although one has to allow for the fact that there may be people with undiagnosed conditions who were uninsured in that group.
A higher percentage of people with multiple chronic conditions had public coverage (about 19.2 percent), which reflects disability coverage in Medicare and Medicaid. Yet two-thirds of people who had three or more chronic conditions had employer-sponsored private insurance.
Some differences by income were observed. A larger percentage of people with multiple chronic conditions were found in the lowest income group, people who are going to be eligible for Medicaid coverage in the ACA. But the differences in family income by health condition are not tremendous.
Cunningham next analyzed the data on out-of-pocket spending by the entire family for both out-of-pocket premiums and out-of-pocket services by family income and health condition. He found that out-of-pocket spending did tend to be higher for people with multiple chronic conditions but the differences were much larger for spending related to services. That reflects the fact that premiums tend to be more predictable. Also, more people are going to be in group coverage, for which there is less variation in the rates, whereas the out-of-pocket spending for services is less predictable.
A much larger percentage of people with multiple chronic conditions tended to have out-of-pocket spending that was greater than 10 percent of their income compared with people with even two chronic conditions, and certainly more than people with acute or no conditions.
One conclusion that can be drawn from these findings is that there tends to be a pretty systematic break for people with three or more chronic conditions compared with people with fewer chronic conditions, resulting in a level of seriousness, or intensity of service use, that leads to greater out-of-pocket spending.
Trends in High Financial Burden
Analysis of MEPS data for 2001, 2006, and 2008 shows that, across all condition categories, between 2001 and 2006 there was an increase in the percentage of people with high financial burden, defined as total out-of-pocket spending greater than 10 percent of family income. That reflects real incomes, adjusted for inflation, basically staying flat for that time period, as well as increases in out-of-pocket costs for both premiums and services. After 2006, however, there appears to be a leveling off for most people of out-of-pocket spending relative to income. For people with three or more chronic conditions, spending actually decreased, returning to 2001 levels.
Most of this trend appears to be related to a decrease in out-of-pocket spending for services; for people with multiple chronic conditions, it is related to a decrease in out-of-pocket prescription drug spending. Other data indicate that, during this time period, there has been a marked shift from brand name use to generic use of prescription drugs. So that could be
accounting for some of the decline. It could also be related to a decrease in demand for medical care related to the recession.
Looking at differences in high burden levels by income, as one would expect, the percentage with high burden was much higher for people with low incomes compared with high incomes, and the gap was actually wider for people with multiple chronic conditions.
Data from the 2-year MEPS panel provides an idea of the persistence of high financial burden. These data show that people with multiple chronic conditions were more likely to have high financial burden over the 2 years. The high burden is more likely to be persistent for people with multiple chronic conditions. This is an important finding, because persistent high burden can lead to accumulated medical debt that can last for years and can have a multiplier effect on the medical cost that one faces in any given year. For example, families might be better able to absorb the costs if they have a one-time medical event that results in substantial medical costs. But if that happens again and again because of a chronic condition, that can become financially burdensome over time.
Problems Paying Medical Bills
Analysis of data from the 2007 Health Tracking Household Survey conducted at the Center for Studying Health Systems Change shows that a higher percentage of people with chronic conditions tended to report more problems paying medical bills, probably due in part to the persistence of high medical cost burdens as well as higher debt levels. A total of 29.9 percent of people with chronic conditions reported problems paying medical bills, compared with 18.5 percent of people with no chronic conditions. What is interesting to note is that higher rates of medical bill problems among persons with chronic conditions was not just the result of higher overall spending. Even people with chronic conditions who had relatively low levels of out-of-pocket spending reported more problems paying medical bills, compared with people with no chronic conditions who reported similarly low levels of spending.
Cunningham concluded that these findings suggest that there is something fundamentally different about having chronic conditions, which should be taken into account when setting affordability thresholds. There is evidence that people with chronic conditions do have lower affordability thresholds than other people. That is especially the case for people with multiple chronic conditions. At similar levels of out-of-pocket spending relative to income, people with chronic conditions reported more problems
paying medical bills. This probably reflects the fact that people with chronic conditions are more likely to have persistently high financial burdens, and that contributes to higher accumulated levels of medical debt that stay around for more than just a year.
There would clearly be a lot of practical issues with implementation. Should all chronic conditions be used, or just a select group? Also, some people with chronic conditions are more likely to choose expensive plans that have higher premiums, in order to cover their expenses. Some measure of affordability needs to be incorporated in data collection. It is important, at least from the standpoint of testing different affordability thresholds, to get some real-world experience from people about whether they find certain levels of out-of-pocket spending to be affordable or not.
TRENDS IN INSURANCE COVERAGE AND THEIR IMPACT ON MEDICAL OUT-OF-POCKET EXPENSES
Gary Claxton (Kaiser Family Foundation) focused mostly on employer-sponsored health insurance and some on nongroup health insurance, where the out-of-pocket expense risks are and how that has changed or not changed over time. He explained that most of the information is from a survey conducted at the Kaiser Family Foundation with the Health Research and Education Trust, but also some data from MEPS, looking at group and nongroup out-of-pocket shares.
Health Insurance Characteristics
He opened his presentation with two observations about what we know and don’t know about health insurance, and its adequacy, and our ability to characterize whether or not someone has good health insurance, and what that means to out-of-pocket risk.
His first point was that there is a lot that is not known about how good health insurance is. Surveys can provide information about deductibles and out-of-pocket maximums, but they cannot keep people long enough on the phone to ask about limits on rehabilitation services, and whether or not biological drugs are covered at certain levels, whether or not they are part of the out-of-pocket maximum, or all the other things that lead to high exposure to out-of-pocket risk, even when people have what they think is good health insurance.
His second point is that the kind of insurance is as important as any trends in terms of out-of-pocket risks. Small-group health insurance plans are different from large-group plans, and they are both different from nongroup health insurance in terms of out-of-pocket burdens and protections.
Claxton observed that premiums are going up a lot faster than either earnings or inflation, and that has been going on for a long time. It is interesting to note that worker contributions track the increase in premiums. So over time workers have been paying the same share of the premium for employer-sponsored health insurance. Last year, for the first time, the shares for both single and family coverage increased, moving from 16 to 18 percent for single and from 28 to 30 percent for family coverage.
Difference Between Small-Firm and Large-Firm Health Insurance
A small firm is defined as one with 3-199 employees. Listening to a lot of the debate in Washington, he said, one would think that small firms pay more for health insurance and that their health insurance premiums are going up faster. Neither of those statements is actually true. What happens is that workers in small firms pay more for their family coverage, and small firms have higher deductibles. But in general small-firm health insurance is somewhat cheaper. It is about $700 cheaper now. There seems to be a trend in that direction: for family coverage, small firms seem to be moving away from large firms and having less comprehensive coverage.
Claxton reported that Kaiser Family Foundation started conducting a survey with the Health Research and Education Trust in 1999 to look at the dollar amount of worker contributions over time for single and family contributions by firm size. They found that, for single coverage, small and large firms were much the same in terms of what workers had to contribute, and it went up about the same amount over time. The big difference is for family coverage: covered workers would have to pay much more for family coverage in small firms than they do in large firms—about a $1,000 difference in the past year. In terms of the share of the premium, the difference for single coverage between small firms and large firms is reversed. Covered workers in small firms on average paid a smaller share of the premium for single coverage, and they paid a much larger share of the premium for family coverage. Part of this has to do with the fact that there are still a number of workers in small firms who have to pay nothing for single coverage. One might guess that this comes from the fact that some insurers require that a certain percentage of workers to enroll before a small firm can get coverage. Small firms are less likely to have Section 125 plans that allow workers to pay their contribution with pretax income. So just covering 100 percent of the premium is a way to get that enrollment.
However, workers in small firms paid on average 35 percent of the cost of family coverage. A family policy was close to $15,000, not a trivial amount. It is therefore not surprising that workers in small firms are much less likely to be enrolled in family coverage than workers in large firms.
Focusing on the tail of the distribution, the percentage of covered workers in small firms and large firms who had to contribute at least 50 percent of the premium, they found that for single coverage, it was not very common. For family coverage in large firms, it was also not very common. For family coverage in small firms, around 30 to almost 36 percent of covered workers in small firms contributed at least 50 percent of premium. That helps explain why there is a lower percentage of workers taking family coverage in small firms.
Looking at the other end, that is, the percentage of covered workers who contributed 10 percent or less of the premiums, in 1999, at the start of the survey, a substantial portion of workers in small firms contributed nothing for their health insurance. That has actually been going down over time, as it has for the other categories as well.
He next focused on cost-sharing. One of the things that is going on with employer-sponsored health insurance is the movement to consumer-funded, consumer-driven plans. In his survey, these are defined as plans that have a deductible of at least $1,000 and can be matched with a savings account. So they are either a health savings account (HSA) qualified plan, or they are a high-deductible plan with a health reimbursement arrangement offered by the employer. Enrollment has grown recently in those plans, and it is going to go up again this year. These plans provide workers with higher deductibles. Higher deductibles are usually matched with somewhat higher out-of-pocket maximums. On the plus side for covered workers, in one of these plans, the employer may make a contribution toward an individual’s savings account. That is by definition true in a health reimbursement arrangement, because only the employer can contribute to those. For workers who are in an HSA-qualified plan, about 60 percent are in plans in which the employer makes a contribution toward their HSA, but not all workers are in that situation.
One other advantage in terms of out-of-pocket costs that comes from the HSA-qualified plans is that the out-of-pocket maximums have to be genuine maximums. That means that all covered spending actually has to count toward the plan’s out-of-pocket maximum. The increase in deductibles is not entirely about people going to consumer-driven plans. The share of workers with plans with a deductible of at least $1,000 has been steadily rising: it is almost half of covered workers in small firms and about a quarter of workers overall. About 20 percent of covered workers in small firms are in a plan with at least a $2,000 deductible, comprising about 10 percent of overall workers. Out-of-pocket maximums are also going up, and this is across all different health plans for people who have an out-of-pocket maximum. The distribution of people with single coverage who are in a plan that has an out-of-pocket maximum of at least $3,000 has
gone up from 22 to 31 percent from 2007 to 2010, and it is going to go up again this year.
Next, using data from MEPS, Claxton described out-of-pocket shares for people with nongroup insurance and group insurance over time. They found that the people with nongroup insurance spend more than half of their total health spending out-of-pocket. That number is pretty persistent. It comes down a little bit over time, but generally it is substantially different than that for people with employer-sponsored insurance, and that is true for both the mean and the median.
He commented on some of the discussion in the earlier session. Work with the American Cancer Society shows that the people who run into trouble are those who have limits on the amount of spending that their policy covers, for example, for radiation. Or they have no real out-of-pocket maximum for drug coverage, and they need biological drugs. Although these situations are fairly rare, their effects can be very large. One does not know about these limits when buying a policy, because this is really a fine point. It is not that uncommon to see a limit on rehabilitation of $3,000 annually in a policy, which may not be adequate if, for example, one has a stroke.
On another point, there was a lot of talk about actuarial value. Actuarial value can mean many things, and it may or may not be related to an actual scope or breadth of benefits in a benefit package. It can be just the percentage of whatever is covered that is paid for. So if the package covers only hospitalization and it pays for all hospitalization, it has an actuarial value of 100 percent.
To relate it to a broad benefit package, for example, one can talk about what percentage of all spending this insurance would cover. But cutting out services for relatively rare events would not affect the actuarial value very much at all. So, for example, one could take out all spending for biological drugs and change the actuarial value 1 or 2 percent. But for the people who need those drugs, that would wipe them out. So it is very hard to actually characterize these policies as protective, if the topic is the out-of-pocket expenditures that people have for catastrophic risk.
Health reform may or may not help with some of these things. There will be an essential benefit package, potentially, but how much it will deal with the scope and duration limits that insurers are allowed in the benefit package is unknown. Also, large employers and self-funded employers are not subject to the essential benefit package.
James Ziliak (University of Kentucky Center for Poverty Research), session discussant, organized his discussion into three unifying themes across the three presentations:
- defining the notion of high burden
- measuring cross-state cost-of-living differences (this applies mostly to the first presentation)
- whether thresholds should vary for different population groups
Ziliak remarked that it was not clear to him why there are different definitions of high medical out-of-pocket spending, depending on income status. In Collins’s presentation, out-of-pocket spending greater than 5 percent of income was considered a high burden for those with annual incomes less than 200 percent of the federal poverty line, and greater than 10 percent of annual income for those above 200 percent of the poverty line. Cunningham’s talk defined burden mainly at 10 percent, except once when he chose a 9.5 percent cutoff for out-of-pocket premiums.
Ziliak thought that it would be useful to choose one number. The question is, should that cutoff of high burden percentage be an endogenous function of the person’s actual spending behavior, or should it be a fixed number closer to the median? Also related to that, high out-of-pocket spending is somewhat of a Southern problem. Using the official Census Bureau definition and state-level poverty rates for 2008, by and large poverty is concentrated in the South at the state level. So choosing a 5 percent cutoff for those living below 200 percent of the poverty line means picking up a higher percentage of people in the South. Also, poverty is known to be correlated with poor health outcomes and is probably also correlated with out-of-pocket spending and lack of insurance.
But to make cross-state comparisons without confounding where the poor live, per se, with how medical out-of-pocket spending is occurring as a fraction of income, one might want to choose one particular cutoff, not different thresholds based on income status. Ziliak suggested using a compromise number between 5 and 10 percent, say 7.5 percent. Choosing a fixed line might improve cross-state comparisons.
Given the salience of the measure of medical care economic risk to the Supplemental Poverty Measure (SPM), perhaps using the same measure of cost of living that is currently being adopted by the Census Bureau for the SPM would potentially improve compatibility between some notion of medical care cost measure and the SPM.
He reported on a conference in April 2011, cohosted by his center with the Census Bureau and Brookings Institution, called Cost of Living and the Supplemental Poverty Measure. Its key recommendation to the Census Bureau was a slight modification of what it is currently doing, which is that the adjustment for geographic housing price differences should be based on quality-adjusted rental costs, not making any adjustments for ownership free and clear or ownership at all. That is readily available data in the American Community Survey (ACS). Given the size of the ACS, it is
an effective way that will allow the capture of rural as well as urban parts of the country.
The last issue is, should thresholds vary across population groups? Yes, is Ziliak’s answer, but then the real question is how and for whom. Cunningham presented important trends in financial burden across the presence and the number of chronic conditions. Claxton presented interesting trends based on firm size and ownership, looking more at family versus single type firms. If the measure of medical care risk is to be prospective, then it suggests that a model is going to need to incorporate something about chronic conditions, firm size, and perhaps self-employment status, to address what type of economic risk an individual is going to face.
Ziliak then spoke about some work that he has contributed to in the last few years, which is the literature called the value of statistical life, which is designed to try to quantify the money-risk trade-off that individuals face. Sometimes people ask questions such as, How much are you willing to pay in order to avoid a 1 in 10,000 increase in the probability of a fatal injury on the job? That is a classic example of the value of statistical life. It suggests that in thinking about risks and what people face and thinking about thresholds and prospective medical care risks, employment status might be a key demographic variable for breaking out the population, because individuals face substantially different risks of on-the-job injury and fatality, depending on their industry and occupation. The Bureau of Labor Statistics collects these data, the Census of Fatal Occupations in Industries. It is broken down by detailed industry and occupation. It could be easily merged into a data set like the CPS, which also collects industry and occupation.
One of the challenges in thinking about the number of thresholds is to know the optimal number one wants to construct for a medical care index. So if one breaks down by employment versus nonemployment and chronic condition versus no chronic condition, guidance is needed, especially for three or more chronic conditions. Perhaps three or more chronic conditions could be an important criterion on which to split. In the context of Claxton’s presentation, he said, firm size seems to be potentially important in terms of demographic groups. However, the CPS does not collect data on chronic conditions.
FLOOR DISCUSSION AND COMMENTS
Participants expressed their views on the various issues flowing from the presentations.
Collins commented that Ziliak’s suggestion of using a 7.5 percent threshold is a good compromise, but one concern might be whether or not it would become an accepted threshold for policy makers, or whether it
would be an acceptable level of expenditures for people, even those in the CHIP program, for example.
The RAND study also used this lower threshold to reflect income that barely covered essentials and also lacked assets and savings among lower income families, she said. Clearly the ACA incorporates a sliding scale measure of both premium affordability standards and cost-sharing.
On the issue of what the affordability threshold should be, Cunningham observed that he did not think that anybody really knows from a strictly empirical basis at which point health care becomes affordable or unaffordable. That has been defined in a lot of different ways.
A lot of the work has used a more normative approach, looking at a moderate- to high-income group to see what their spending is. This means, in effect, looking at various percentile distributions that should be defined as what is affordable. He said he is not completely comfortable with that because of the way a lot of out-of-pocket spending works in health care, especially if there are unexpected costs, such as a need to go to the hospital or a need to get a procedure done. People cannot always adjust their spending based on what they can afford. They get the care that they need at the time they need it, and they deal with the bills later—and that is when medical debt issues come up. There is also the issue of whether affordability should be based on necessary versus unnecessary costs. And nobody really knows exactly what those are, either.
There is also some justification for using a lower threshold for low income, he said. In fact, in the CHIP program, a 5 percent threshold is used; any out-of-pocket costs cannot exceed 5 percent—so there is some justification or rationale. At this time, the threshold that is used, whether it is 5 percent, 10 percent, or some in between, is fairly arbitrary.
Steven Cohen (Agency for Healthcare Research and Quality) directed his comments to Cunningham. He found the presentation quite informative, showing what one would expect in terms of moving from no conditions to acute conditions to multiple chronic conditions and using family income and family out-of-pocket expenditures. He suggested that individuals with no conditions probably are in families with people with three or more conditions, if one did the analysis in which the context was at the person level, so they are carrying that risk. Likewise, people at the high end with three or more conditions are carrying individuals with no conditions. There would be a more dramatic step up if one did the analysis by family size. He asked Cunningham if he did that, what kind of tail ends of the burden estimates would he see.
Cunningham responded that he did not analyze the data by family size. He did separate it out by self only versus family plan to get an idea of what the individual spending is. But the family spending is driven by people with health problems, and it is not the case that people with health problems
are clustered in individual families. Sometimes one does see the correlation, but the family spending is driven by people with chronic conditions. The idea of affordability is also part of the question—whether it should be done at the family level, meaning that even people without health problems are carrying the burden. Even the healthy people in families with others who spend a lot on health care are going to be affected by that spending to the extent that it affects decisions regarding other spending that goes on in the household.
Claxton added that, even forgetting about chronic conditions, if one looks at families, the person with the highest spending on average contributes 70 percent of the family’s spending generally. It is not just families with chronic conditions; it is generally true that the person who spends the most really drives the family spending.
Emmett Keeler (RAND) stated that, as a general point, the generosity of insurance is a way of transforming medical expenses into premiums. With generous insurance, one pays a high premium and has low out-of-pocket expenses. Because it is a transfer, one would always want to include the premiums as part of the research on burden and affordability. He also asked why, in Claxton’s presentation, small firms would give up the tax benefits of providing all the premium. He said it must relate to certain market factors that he does not understand.
Claxton responded that focus groups he has conducted with small firms indicate that the average wage in small firms is much lower than the average wage in larger firms. So there is probably less tax benefit there. And given the number of two-worker families now, small firms like it when the family members work at large firms. They feel they need to contribute for their own workers, but small firms are much less likely to do so for the family members. Keeler agreed, pointing out that the same situation occurs in the military health care system.
Claxton added that, based on some focus groups and some additional research done last year, they found that with very small firms, those differences are even more pronounced. Many times the owners of very small firms—with fewer than 20 workers, and in some cases fewer than 10—said that they contributed a certain percentage toward single coverage, which could be 50 or 100 percent, but they contributed nothing in addition for family coverage.
Thesia Garner expressed concern about different references to measures. The first session was primarily focused on the medical care risk index. And this session was identified as focusing on thresholds. Ziliak talked about adjusting thresholds, and Cunningham spoke about an affordability threshold or index. She said that they were talking about different things: one is more of an ex ante and the other is more of an ex post measure. She said she was not sure what this affordability measure is because it is not
the same thing as the Supplemental Poverty Measure, and it does not seem to be the same as a measure of medical care economic risk.
Wilhelmine Miller (NORC at the University of Chicago) made a similar comment, stating that it seemed like the entire session addressed a range of factors that affect the likelihood of incurring high medical care economic risk, some of which are perhaps more appropriate for applying in a policy analytic sense, and others not. She said that although Ziliak treated them all as threshold issues, she was not sure, for example, that labor category is a threshold issue, even though job type has a lot to do with the ultimate risk of both expenses and debt. She wanted to hear from the presenters, who talked about everything from geographic variation to risk based on health condition, as to what is most salient.
Cunningham responded that although they were using different terminology, some of the issues are still the same in terms of what should be going into some kind of medical care risk index. For example, whether one uses chronic conditions or some other measure of people at risk for high spending, the issue is still the same. What the presenters are grappling with is whether there is something unique to people with chronic conditions or low-income people above and beyond what they spend or what their risk is for spending on an annual basis that should be considered.
In his talk, he said he pointed out that, for people with chronic conditions, their risk is not only for spending in any given year. It is also the risk for having the spending persist over a period of time, and that is qualitatively different from an annualized measure. He urged people not to get caught up in the differences in terminology, because the issues are still relevant to how should the index be computed, how one should do it, and whether there should be different considerations for different groups.
Garner questioned whether he was talking about a supplemental poverty threshold the way Ziliak was, when he said that people make spending decisions on health care based on their other spending needs. She said she thought that he is not just thinking about medical care economic risk, he is also saying that people have constraints in their household, which is exactly what the 1995 NRC report says and also what the Interagency Technical Working Group document says, that people make decisions interdependently. Medical expenditures are not independent from food, clothing, shelter, utilities, and the like. Or are they?
Garner continued the fact is that people with health insurance have greater flexibility when it comes to making those decisions. If they have health insurance, their decisions relative to food, clothing, shelter, and utilities are different than if they do not have health insurance. Cunningham responded that it is not an either/or situation; the distinction is between predictable and unexpected health care spending. With premiums, people are sharing the cost of health care with other people.
Employer-sponsored coverage comes out of one’s paycheck. It is a regular expense that one gets used to and doesn’t even notice. But because of the increases in cost-sharing that have been occurring, people may start to have unexpected expenses. This is true even for people with private insurance. Maybe they have a high deductible; maybe the expense is not covered by their plan. And that is when their decisions are affected by other spending in the household.
Cathy Schoen (Commonwealth Fund) raised the question of whether the work going forward on tracking reform progress should focus on prospective risk or look at what people did spend.
She asked what people want to do with this index. Tracking would show whether the reforms put on the books have converted more of the risk, moving it from risk into a premium, and making it more predictable.
In her view, a measure is needed to indicate whether policy is moving in the right direction, staying about the same, or getting worse over time. The current SPM just subtracts medical care spending and says who is poor now—that is already on the books. Did they become poor just because of medical care?
Claxton responded that depending on some of the health reform decisions and what packages people get—particularly in the nongroup insurance market, in which the risk is the largest because the policies are the shabbiest—one might be able to say something prospectively about the insurance if the essential benefit package does not have a lot of limits.
If the package has a lot of limits, then one wants to track what people actually spend in order to figure out where they are spending their money and why. There is supposed to be an out-of-pocket limit of $5,000, $6,000, or whatever is decided upon.
For people who do not get a lot of cost-sharing subsidies, the deductibles are going to be in the several-thousand-dollar range at least, the out-of-pocket maximums will be nontrivial, and there may be other limits. So people with chronic conditions are still going to spend out-of-pocket. One hope is that fewer people will be spending very large amounts because they are hospitalized.
Those out-of-pocket limits apply to in-network care, he continued. And for people who are hospitalized, finding their way to an in-network provider all the way through their hospitalization is very difficult. The hospital doesn’t always wake you up from surgery to ask you about your network, for example.
Pamela Short mentioned two different takes on what represents a high burden of out-of-pocket expenses. One is an absolute standard of, say, 10 percent or 7.5 percent of income, which implies that, even at 400 or 500 percent of the poverty line, it is a bad thing, trying to avoid losing 10 percent of your consumption, basically on a much higher base. She said that
should be thoughtfully contrasted with a different approach, which would use a relative threshold or a relative burden as opposed to a more absolute standard, which would say that a high burden is when out-of-pocket spending on medical care and/or premiums drops you below that subsistence level. That approach is more embedded in the SPM.